JPH096549A - Laminated disk array device - Google Patents

Abstract

PURPOSE: To constitute a disk array with RAID tower obtained by integrally laminating LSIs for performing an interface control between a display array controller and drives, directly loading and laminating the drives on the LSI. CONSTITUTION: The LSI 4 for controlling the interface between the disk array controller and drives 2 is integrally laminated, the drives 2 are directly loaded and laminated on the LSI 4 to constitute the disk array with the RAID tower. Consequenly an extremel miniaturized disk array can be attained, connecting in a short distace can be attained by pin connection without using a cable and the control of data transfer speed and the display array can quickly be executed by high performance.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk subsystem device capable of highly reliable and highly available input / output operation.

[0002]

2. Description of the Related Art In a current computer system, data required by a CPU is stored in a secondary storage device, and data is written to and read from the secondary storage device when the CPU requires it. Has become. A non-volatile storage medium is generally used as the secondary storage device, and typical examples thereof include a magnetic disk device and an optical disk. Hereinafter, the magnetic disk device will be referred to as a drive. Currently, the drive used in the secondary storage device is 3.5 inches, and the interface is SCSI.

On the other hand, Japanese Patent Laid-Open No. 3-108178 discloses a method of directly mounting a drive having a small diameter and a small size on a substrate on which a semiconductor device is mounted in the same manner as a semiconductor device for the purpose of reducing power consumption of the drive. Disclosure.

In Japanese Patent Laid-Open No. 4-228153, Japanese Patent Laid-Open No.
No. 108178, a method is disclosed in which a drive is directly mounted on a substrate on which a semiconductor device is mounted in the same manner as a semiconductor device, and a disk array is configured by a plurality of drives mounted on the substrate.

Further, in Japanese Patent Laid-Open No. 4-276375,
In order to realize a small-sized drive with low power consumption, a magnetic disk device in which a connector for control signals and data signals with an external system and a power supply terminal are provided on a circuit board of the drive is disclosed.

Japanese Patent Application No. 6-156975 discloses a method of mounting a plurality of drives and a control mechanism for controlling these drives on the same substrate.

[0007] These prior arts disclose only a method of mounting a drive on a substrate, and do not disclose a method of stacking a plurality of drives and forming a disk array by the stacked drive group. Also,
LS that controls the disk array when stacking drives
It does not disclose a method of stacking I in the same manner and stacking the LSI and the drive to form a disk array.

[0008]

[Patent Document 1] Japanese Patent Application Laid-Open No. 4-228
In Japanese Patent Laid-Open No. 153/153, similarly to Japanese Patent Laid-Open No. 3-108178, a drive is directly mounted on a substrate on which a semiconductor device is mounted similarly to a semiconductor device, and a disk array is formed by a plurality of drives mounted on the substrate. The method is disclosed. However, according to this method, the disk array can be constructed only by the number of drives on the board, and the number of drives on the board is limited in area, and many drives cannot be arranged. If you want to place many drives, you have to use a large board.

An object of the present invention is to stack a control LSI for controlling a drive and a disk array of the drive, and configure the disk array with the stacked LSI and the drive. Another object is to make it possible to freely set the configuration of the disk array thus laminated. Another object is to bring the disk array control device and the interface device for the drive, which are conventionally placed away from the drive, into close contact with the drive.

[0010]

According to the present invention, a drive and a control LSI for controlling the interface of the drive and a disk array are stacked, and the stacked LSI and the drive constitute a disk array. A group of this stacked LSI and drive is called a RAID tower, and this RAID tower unit becomes a disk array.

[0011]

In the past, the disk array could be constructed only by the drives on the board, but by using the present invention, the disk array can be constructed by the three-dimensionally mounted drives. It is possible to configure the disk array by further increasing the packaging density. In addition, the interface control of the disk array and drive can be performed with high performance by bringing the disk array control device and the interface control device of the drive, which were conventionally placed away from the drive, into close contact with the drive. Become.

[0012]

Embodiments of the present invention will be described below with reference to the drawings.

(First Embodiment) (1) Outline of Configuration FIG. 1 is an external view showing a mounted state of a RAID tower 1 according to a first embodiment of the present invention. A microprocessor and drive for RAID control of the drive are mounted on each board (board) 3.
LSI that performs both interface control of 2 (hereinafter RA
A CPU 5 and a memory 6 are mounted on the board in the same manner as the drive 2 (called an ID controller). A plurality of drives 2 are stacked and mounted on each RAID controller 4 to form a RAID tower 1. Board 3 is a motherboard
It is inserted in the connector 5 provided on the board 4 and can be removed from the motherboard 4 by pulling it out.

FIG. 7 shows an internal configuration diagram of the drive 2. The drive 2 has a magnetic disk 100 mounted on a main substrate 110 by bearings 113.
Data is read or written by the head 101 upward. The head 101 is moved by a voice coil motor 102 and positioned on a target track. The data read from the head 101 is amplified by the read / write amplifier 103 and sent to the read / write circuit on the circuit board 105. In the case of writing, conversely, data is sent from the read / write circuit to the read / write amplifier 103, and the data is written from the head 101 onto the magnetic disk 100. The operation of the drive 2 is the same as that of a general drive. Next, the features of the drive 2 of the present invention will be described. The magnetic disk 100 and the head 101 of the drive 2 are the housing 11
It is covered by 2 and protected from dust. The drive hole 11 is formed in the housing 112, and the drive pin 12 and the board pin 10 are inserted into the drive hole pipe 104 on the main board 110 through the drive hole 11 and connected to the circuit board 105. The housing 112 and the main board 110 are screwed and connected by a fixing screw 111.

FIG. 8 shows a sectional view of the drive 2. As you can see from this figure, drive hole pipe 104
Penetrates the housing 112 through the drive hole 11 and is connected to the circuit board 105. The drive pins 12 and the board pins 10 inserted into the drive holes 11 come into contact with the drive hole pipes 104, so that data can be exchanged with the read / write circuit 106 on the circuit board 105. The drive pin 12 is also connected to the circuit board 105 similarly to the drive hole pipe 104, and data can be exchanged with the read / write circuit 106 on the circuit board 105. The drive pin 12 and the board pin 10 are fixed by a drive hole pipe 104 to fix the connections between the stacked drives 2 and between the boards 3 and the drives 2.
As a result, stacked mounting of the drive 2 and direct connection with the board 3 are possible.

According to the present invention, the drive pin 12 or the board pin 10 is thus inserted into the drive 2. The length of drive pin 12 and board pin 10 is 5 ± 2mm
And the height of the drive 2 is also 5 ± 2mm. If the diameter of the magnetic disk 100 is 1 inch, the width of the drive 2 is 30 ± 4 mm and the length is 40 ± 5 mm.

FIG. 2 shows a method of constructing the RAID tower 1. As shown in FIG. 2, the board 3 has a RAID controller 4
A hole (board hole) 13 for connecting with is formed.

The RAID controller 4 has a drive 2 on the top surface.
A hole (RAID controller hole) 14 into which the board pin 12 is inserted is opened, and a pin (RAID controller pin) 15 to be inserted into the board 3 is provided on the lower surface. By inserting the drive pin 12 of the drive 2 into the hole 14 of the RAID controller 4, the drive 2 and the RAID controller 4 are connected. Signals and data of the SCSI interface flow through the RAID controller hole 14 and the drive pin 12. Similarly to the connection between the RAID controller 4 and the drive 2, the drives 2 are also connected. Each drive 2
Drive hole 11 into which drive pin 12 of another drive 2 is inserted on the upper surface and drive pin 1 on the lower surface
2 is out. When stacking between drives 2, it is possible to connect two drives 2 by inserting the drive pin 12 of one drive into the drive hole 11 of the other drive. It will be possible. Signals and data of the SCSI interface flow through the drive hole 11 and the drive pin 12 of each drive 2.

In this way, the drive pin 12 is inserted into the drive hole 11 and the drive pin 12 of each drive is inserted.
Signals and data are exchanged between the drives 2.

In the present invention, the number of drives to be stacked is not limited.

(2) Outline of internal operation FIG. 3 shows a board 3 on which the RAID tower 1 is mounted. Therefore, an outline of the internal operation of the RAID tower 1 will be described with reference to FIG.

The read or write command issued by the CPU 20 is input to the RAID controller 4 of the RAID tower 1 through the MP bus 16. CPU20 to RA
When a command is issued to the ID controller 4, the RAID controller 4 determines whether the command can be accepted.
If it can be accepted, the RAID controller 4 performs command acceptance processing. If it is not accepted because another read or write process is being performed, RA
The ID controller 4 sends an unacceptable response to the CPU 20. The command from the CPU 20 is a data write command, and the write data transferred from the CPU 20 is data # 1.

In the present embodiment, it is assumed that the data length of data written or read from the CPU 20 is always fixed length data of 4 KB. Therefore, the write data # 1 from the CPU 20 is also 4 KB. The present invention is not limited to this data length.

The write data # 1 from the CPU 20 is
The data is transferred to the RAID tower 1 via the MP bus 16, and the RAID controller 4 of the RAID tower 1 instructs the CPU 20 to transfer the data # 1 to the memory 22 via the MP bus 16. After storing the data # 1 in the memory 22, the RAID controller 4 sends the data to the CPU 20.
Report the completion of storing # 1 in the memory 22. Upon receiving the completion report, the CPU 20 instructs the RAID controller 4 to issue a write request to the drive 2. Upon receiving this instruction, the RAID controller 4 writes the data # 1 to the drive 2 according to the processing procedure of the SCSI interface.

Data is written from the CPU 20 as described above.

When the CPU 20 requests the reading of the data, the data is read from the drive 2 and stored in the memory 22, contrary to the above-mentioned writing process.
It is sent to the CPU 20 via the P bus 16.

Next, this embodiment will be described in more detail. In the following, regarding the address conversion table in (3),
The read and write operations will be described in detail in (4).

(3) Address Conversion Table In this embodiment, a SCSI interface drive is used as the drive 2 constituting the RAID tower 1. Therefore, the CPU 20 issues a write or read command using the conventional interface.

When the operating system (OS) in the CPU 20 instructs the RAID controller 4 of the RAID tower 1 to write data, the OS in the CPU 20 attaches a logical address (data name or data number) to the written data. Transfer to the RAID controller 4. At this time, when the write data is updated in the RAID controller 4, the address table 30 shown in FIG.
From the logical address 31, the tower number 32 which is the address of the RAID tower 1 in which the drive 2 storing this data is mounted, the RAID level 33, the drive number 34 which is the address of the drive 2 and the drive number 34 in the drive 2 The SCSI address 35, which is the physical address where the data is actually stored, the logical address 37 of the parity related to the data to be updated, and the parity drive number 38 where this parity is stored are recognized.

On the other hand, when the write data is new write data, the CPU 20 searches the address table 30 for an entry in which the logical address 31 is not registered, and registers the logical address there. After registering the logical address 31 of the address table 30, the tower number 3 which is the address of the RAID tower 1 in which the drive 2 in which this data is stored is mounted is the same as the case of updating.
2, RAID level 33, drive number 34 which is the address of the drive 2 and SCSI address 35 which is the physical address where the data is actually stored in the drive 2
And the logical address 3 of the parity related to the data to be updated
7 and the parity drive number 38 in which this parity is stored are recognized.

In the address table 30, the drive failure flag 39 is turned on (1) when a failure has occurred in the drive 2, and is turned off (0) in the normal state. The setting of the drive failure flag 39 cannot be executed when the RAID controller 4 issues a read or write request to the drive 2 and retry is performed several times, even if the retry is performed beyond the set number of times. If it cannot be executed, the RAID controller 4 recognizes that a failure has occurred in drive 2 and drives failure flag 3 in address table 30.
Set 9 to on (1).

The address table 30 is stored in an appropriate area in the memory 22 for the RAID controller 4. The address table 30 is automatically read into the memory 22 from a specific drive 2 when the system is powered on. On the other hand, when turning off the power, RA
The ID controller 4 automatically stores the address table in the memory 22 at a predetermined location in the original drive 2.

(4) Data Read and Write Operations Next, RAID using the address table 30 as described above
The data read or write processing by the controller 4 will be described in detail. The RAID controller 4 has a single LSI that contains hardware that performs interface control between a processor that performs disk array control and drives. Shown below is the RAID controller 4
This is an operation by a microprogram of the processor unit for controlling the disk array in the above.

FIG. 5 is based on the address table 30 of FIG.
It shows a state in which data and parity are actually stored. Specifically, data # 1 is stored in the SCSI address 35 of SD # 1 of drive # 1, and parity # 2 is stored in SD # 2.
Data # 3 is stored in SD # 3. SCSI address 3
Reference numeral 5 indicates a physical address in the drive 2, which indicates a cylinder address, a head number, and a data number in a track. In Figure 5, drives # 1, 2, 3, 4
Parity # 1 is created from data # 1, 4, 7 of SD # 1, data # 5 of SD # 2 of drives # 1, 2, 3, 4
Parity # 2 is created from 8 and 11, drive # 1,
Parity # 3 is created from the data # 3, 9, 12 of SD # 3 of 2, 3, and 4, and the parity is drive # 1, 2, 3,
Since it is distributed over 4, it is RAID 5.

The write processing method will be described with reference to the flowchart of FIG.

In FIG. 3, it is assumed that the CPU 20 first issues a read or write command to the RAID controller 4 (step 41).

The RAID controller 4 identifies the tower number 32, RAID level 33, drive number 34, drive failure flag 39, and SCSI address 35 from the logical address 21 by using the address table 30 in the memory 22 (step 42). Next, it is judged whether the request issued from the CPU 20 to the RAID controller 4 is a read request or a write request (step 43).

In the case of a write request, the parity logical address 37 and the parity drive number 38 are specified from the logical address 21 by the address table 30 (step 44). RAID controller 4 is an SC that stores data
A read request for the old data and the old parity is issued to the SCSI address 35 in which the old parity is stored, which is specified from the old data and the parity logical address 37 stored in the SI address 35 (step 45). The drive 2 that issued the read request waits for seek and rotation for the SCSI address 35, and the old data and old parity can be read, and the old data and old parity can be read from the person who can use the SCSI bus. Is read (step 46). When the reading of old data and old parity to the RAID controller 4 is completed, the relevant RAID controller 4
In step S4, the exclusive OR is performed on the old data, the old parity, and the new data to create a new parity (step 4).
7). Since the new data is stored in the memory 22 from the CPU 20 in advance, when this exclusive OR is performed, the RA
The ID controller 4 reads the new data from the memory 22. Next, the RAID controller 4 issues a write request for new data and new parity to be written to the SCSI address 35 of the drive 2 (step 4).
8). In the relevant drive 2, the relevant SCSI address 3
Seek to 5 and wait for rotation (step 49), and new data and new parity can be written.
New data and new parity are written from the person who can use the bus (step 50).

On the other hand, as a result of judging the request from the CPU 20 in step 43, if the request is a read request, a read request is issued to the SCSI address 35 of the drive 2 (step 51). In the drive 2 to which the read request is issued, seek and rotation waiting are performed for the SCSI address 35, and the read is performed when the data can be read from the person who can use the SCSI bus (step 52).

Next, a data recovery method when a failure occurs in the drive 2 will be described with reference to FIG. Drive # 1 in the RAID 5 disk array as in this embodiment
If a failure occurs in the drive # 1 and the data in the drive # 1 is destroyed, the data in the drive # 1 can be recovered from the drives # 2, 3, and 4. Specifically, the data # 1 stored at the address of SD # 1 of drive # 1 is
Data stored in the SD # 1 address of drive # 2
Recovery is performed by taking the exclusive OR of the data # 7 stored at the address of SD # 1 of drive # 3 and the parity # 1 stored at the address of SD # 1 of drive # 4. Is possible. Similarly, the data # 3 stored at the address of the SD # 3 of the drive # 1 is the drive # 2.
Parity # 3 stored at the SD # 3 address of the drive # 3, Data # 9 stored at the SD # 3 address of the drive # 3
It is possible to recover by taking the exclusive OR of the data # 12 stored at the address of SD # 3 of drive # 4. In this way, when a failure occurs in the drive 2, the data in the failed drive 2 can be recovered.

According to the present invention, the RAID controller 4 for controlling the disk array and the interface with the drive and the drive 2 are laminated as shown in FIG. The feature is that a very small disk array can be realized because it is not necessary to connect the device and the drive 2 with a cable. Also, instead of connecting with a cable, it is possible to connect over a short distance by using pins, and it has become possible to perform data transfer speed and disk array control at high speed and with high performance.

(Embodiment 2) In this embodiment, a plurality of RAID towers 1 are connected to one RAID controller 4.

FIG. 9 shows the overall configuration of this embodiment. One RAID controller 4 is provided with a RAID tower (1) 1
The two RAID towers 1 of the RAID tower (2) 1 are connected. In the example of FIG. 9, the RAID tower (1) 1 is the same as that of the first embodiment, and the RAID tower (2) 1 has a different configuration from the RAID tower (1) 1. In other words, RAID tower (1) 1 is RAID5 with four drives 2, but RAID tower (2)
RAID 1 is a RAID 1 that uses two drives for duplication. Further, in FIG. 9, the RAID tower (3) 1 and the RAID tower (4) 1 are also connected on one RAID controller 4, but the RAID tower (3) 1 and the RAID tower (4) 1 are the same as those in the first embodiment. RAID 5 is made up of 4 drives 2 as in. As described above, the present embodiment is characterized in that a plurality of RAID towers 1 are connected to one RAID controller 4, and the configuration of each RAID tower 1 can be freely set.

FIG. 10 shows a method of actually connecting a plurality of RAID towers 1 to the RAID controller 4. Example 1
Similarly, the board 3 is provided with a hole (board hole) 13 for connecting to the RAID controller 4. RA
The ID controller 4 has a board pin 1 of the drive 2 on the upper surface.
Hole into which 2 is inserted (RAID controller hole) 1
4 is opened so that 2 drives 2 can be inserted, and pins (RAID controller pins) 1 to be inserted into the board 3 on the lower surface
5 is out. In the present embodiment, two RAID towers 1 are connected to one RAID controller 4, but 3
More than one RAID tower 1 may be connected. The method of connecting the drive 2 to the RAID controller 4 is the same as in the first embodiment. In other words, drive pin 12 of drive 2 to R
The drive 2 and the RAID controller 4 are connected by being inserted into the hole 14 of the AID controller 4. Signals and data of the SCSI interface flow through the RAID controller hole 14 and the drive pin 12. Similarly to the connection between the RAID controller 4 and the drive 2, the drives 2 are also connected. Each drive 2 has a drive hole 11 on the upper surface into which a drive pin 12 of another drive 2 is inserted and a drive pin 12 on the lower surface. When stacking between the drives 2, by inserting the drive pin 12 of one drive into the drive hole 11 of the other,
It is possible to connect two drives 2, and similarly, it is possible to stack a plurality of drives. Signals and data of the SCSI interface flow through the drive hole 11 and the drive pin 12 of each drive 2.

In this way, the drive pin 12 is inserted into the drive hole 11 and the drive pin 12 of each drive is inserted.
Signals and data are exchanged between the drives 2.

(2) Outline of internal operation FIG. 11 shows a board 3 on which the RAID tower 1 is mounted.

The outline of the internal operation of the board 3 is the same as that of the first embodiment. The difference from the first embodiment is that the RAID tower 1 has one set in the first embodiment and therefore the drive path 130 in the RAID tower 1 is one, but in the present embodiment, two sets are provided in one RAID controller 4. RAID towers 1 are connected, and each drive 2 is installed in these RAID towers 1.
Are connected by the drive path 130, the number of the drive paths 130 is two.

Next, this embodiment will be described in more detail. In the following, regarding the address conversion table in (3),
The read and write operations will be described in detail in (4).

(3) Address Conversion Table When the operating system (OS) in the CPU 20 instructs the RAID controller 4 of the RAID tower 1 to write data as in the first embodiment, the OS in the CPU 20:
The write data is given a logical address (data name or data number) and transferred to the RAID controller 4. In this embodiment, when the write data is updated in the RAID controller 4 at this time, the RAID tower 1 in which the drive 2 in which this data is stored is mounted from the logical address 31 by the address table 30 shown in FIG.
It recognizes the tower number 32 which is the address of. When the tower number 32 is RT # 1, the RAID level 3 is the same as in the first embodiment.
3 and the drive number 34 that is the address of the drive 2
The SCSI address 35, which is the physical address where the data is actually stored in the drive 2, the logical address 37 of the parity related to the data to be updated, and the parity drive number 38 where this parity is stored are recognized.

However, if the tower number is RT # 2, RAID
It recognizes the level 33, the drive number 34 that is the address of the drive 2 and the SCSI address 35 that is the physical address in the drive 2 that actually stores data.

On the other hand, when the write data is new write data, the CPU 20 searches the address table 30 for an entry in which the logical address 31 is not registered, and registers the logical address there. If you want to store in RAID5, look for the entry where the logical address 31 is not registered in the area where the tower number 32 is RT # 1 and
If you want to store in RAID1, look for an entry in which the logical address 31 is not registered in the area where the tower number 32 is RT # 2. After registering the logical address 31 of the address table 30, as in the case of updating, the tower number 32
The information in the address table 30 is recognized by.

(4) Data Read and Write Operations Next, RAID using the address table 30 as described above
The data read or write processing by the controller 4 will be described in detail. The following is the operation by the microprogram of the processor unit that controls the disk array in the RAID controller 4.

The write processing method will be described with reference to the flowchart of FIG.

In FIG. 13, first, CPU 20 performs RAID.
It is assumed that a read or write command is issued to the controller 4 (step 132).

The RAID controller 4 recognizes the tower number 32 from the logical address 21 by the address table 30 in the memory 22 (step 133). Tower number 32
From the address table 30 after recognizing the
3, drive number 34, drive failure flag 39, SCSI
The address 35 is specified (step 134). Next, from the RAID level 33 recognized in step 134, it is determined whether the data is subjected to RAID5 processing or RAID1 processing (step 135). If it is a RAID5 process, the same process as in the first embodiment is performed. However, in the case of RAID1 processing, it is determined whether the request issued from the CPU 20 to the RAID controller 4 is a read request or a write request.
(Step 136).

If a write request is made, two drive numbers 34 are specified from the logical address 21 by the address table 30 and a new data write request is issued to the two drives 2 (step 137). In the drive 2 that was issued the write request, the SCSI address 3
A seek and a rotation wait are performed for 5 (step 138).
When the SCSI bus can be used and new data can be written, the new data is written from one of the drives (step 139).

On the other hand, as a result of judging the request from the CPU 20 in step 136, if the request is a read request, a read request is issued to the SCSI address 35 of the drive 2 (step 140). In the drive 2 to which the read request is issued, the SCSI address 35
On the other hand, seek and rotation wait are performed, and when the SCSI bus can be used and the data can be read, the data is read (step 141).

Next, when a failure occurs in the drive 2, in the RAID tower 1 which is processing by RAID5 with the tower number 32 of RT # 1, the recovery processing is performed as in the first embodiment. if,
If the drive 2 fails in the RAID tower 1 that is processed by RAID1 with RT # 2 as RT # 2, the data is read from the normal drive 2 because the data is duplicated in RAID1. You can put it out.

A feature of this embodiment is that a plurality of RAID towers 1 are connected to one RAID controller 4 and a plurality of RAID towers having different configurations and functions can be mounted on one RAID controller 4. .

(Third Embodiment) In the first and second embodiments, the drive bus 130 through which data and commands in the RAID tower 1 can be passed.
Was one. Therefore, in the first embodiment, when a read request for old data and old parity is issued to the drives # 1 and # 2 as shown in FIG.
Even if the EK and rotation waits are completed, the drive bus 130, which is the SCSI bus, is used by the drive # 2, so the drive # 1 must wait until the drive bus 130, which is the SCSI bus, becomes available. Therefore, in this embodiment, as shown in FIG. 15, the drive bus 130 for connecting the RAID controller 4 and each drive 2 is replaced by the drive bus 1
There are two, 150 and drive bus 2 151. As a result, even if one drive bus 1150 is in use, processing can be performed by using the other drive bus 2151. In this embodiment,
2 of drive bus 1 150 and drive bus 2 151
By preparing a drive bus for a book and making it possible to use whichever bus is available, there is no need to wait because the bus cannot be used. FIG. 16 shows the RAID of this embodiment.
RA is an explanatory diagram of how to connect the controller 4 and drive 2.
Two sets of RAID controller upper holes 14 are formed on the ID controller 4, and pins (RAID controller pins) 15 to be inserted into the board 3 are provided on the lower surface. Also in this embodiment, the drive pin 12 of the drive 2 is the same as in Embodiments 1 and 2.
The drive 2 and the RAID controller 4 are connected by inserting into the hole 14 of the RAID controller 4. However, in this embodiment, two sets are prepared. Signals and data of the SCSI interface flow through the RAID controller hole 14 and the drive pin 12. Similar to the connection between the RAID controller 4 and the drive 2, the drives 2 are also connected, but two sets are prepared for each drive 2.
Each drive 2 has two sets of drive holes 11 on its upper surface into which drive pins 12 of another drive 2 are inserted, and two sets of drive pins 12 on its lower surface. When stacking the drives 2, two drives 2 can be connected by inserting the drive pin 12 of one drive into the drive hole 11 of the other drive as in the first and second embodiments. It is possible to stack multiple drives. Signals and data of the SCSI interface flow through the drive hole 11 and the drive pin 12 of each drive 2.

In this way, the drive pin 12 is inserted into the drive hole 11 and the drive pin 12 of each drive is inserted.
Signals and data are exchanged between the drives 2.

[0062]

In the past, the disk array could be constructed only by the drives on the board.
By using the present invention, it is possible to configure a disk array with a three-dimensionally mounted drive, and it is possible to configure a disk array with a higher packaging density per substrate. In addition, the interface control of the disk array and the drive can be performed with high performance by bringing the disk array control device, which was conventionally placed away from the drive, into close contact with the drive.

[Brief description of drawings]

FIG. 1 is a first overall configuration diagram of the present invention.

FIG. 2 is an explanatory diagram of a board and drive connection method according to the first embodiment.

FIG. 3 is an internal configuration diagram of the first embodiment.

FIG. 4 is an explanatory diagram of an address table according to the first embodiment.

FIG. 5 is an explanatory diagram of data storage according to the first embodiment.

FIG. 6 is an explanatory diagram of reading and writing processing according to the first embodiment.

FIG. 7 is a drive configuration diagram of the first embodiment.

FIG. 8 is a drive sectional view of the first embodiment.

FIG. 9 is an overall configuration diagram of a second embodiment.

FIG. 10 is an explanatory diagram of a board and drive connection method according to the second embodiment.

FIG. 11 is an internal configuration diagram of the second embodiment.

FIG. 12 is an explanatory diagram of an address table according to the second embodiment.

FIG. 13 is an explanatory diagram of reading and writing processing according to the second embodiment.

FIG. 14 is a time chart of the third embodiment.

FIG. 15 is a configuration diagram of a third embodiment.

FIG. 16 is an explanatory diagram of a board and drive connection method according to the third embodiment.

[Explanation of symbols]

1 ... RAID tower, 2 ... Drive, 3 ... Board, 4 ... RAID
Controller, 5 ... Connector, 6 ... CPU memory, 7 ...
Motherboard, 10 ... Board pin, 11 ... Drive hole, 12 ... Drive pin, 13 ... Board hole, 14 ...
RAID controller upper hole, 15 ... RAID controller pin, 16 ... MP bus, 20 ... CPU, 22 ... Memory, 30
... address table, 31 ... logical address, 32 ... tower number, 33 ... RAID level, 34 ... drive number, 35
… SCSI address, 37… Parity logical address, 38…
Parity drive number, 39 ... Drive failure flag, 1
00 ... Magnetic disk, 101 ... Head, 102 ... Voice coil motor, 103 ... Read / write amplifier, 10
4 ... Drive hole pipe, 105 ... Circuit board, 106
... Read / write circuit, 110 ... Magnetic disk, 11
1 ... Fixing screw, 112 ... Housing, 113 ... Bearing,
130 ... Drive bus, 150 ... Drive bus 1, 15
1 ... Drive bus 2.

Claims (12)

[Claims] 1. A board having a hole for connecting a RAID controller for controlling disk array control and interface with a drive, a RAID controller having input / output pins, and a disk drive having input / output pins. , The RAID controller pins are inserted into the holes, and the upper surface of the RAID controller has holes into which the drive pins are inserted. The drive pins are inserted into the RAID controller holes and the drive and RAID controller Each drive also has a drive hole on the top surface where the pin of the other drive is inserted, and by inserting the pin of one drive into the hole of the other drive, the two drives are connected and the drive is connected. Stack and stack multiple drives and RAID controller to form a RAID tower. Stacked disk array apparatus, characterized in that. 2. The stacked disk array device according to claim 1, wherein data and commands are transferred through the holes and pins of the RAID controller and the holes and pins of the drive. 3. The stacked disk array device according to claim 1, wherein disk array control is performed by the stacked RAID controllers and drives. 4. The stacked disk array device according to claim 1, wherein the RAID controller is composed of one LSI. 5. A tower number which is an address of the RAID tower, a RAID level, a drive number which is an address of the drive, a SCSI address which is a physical address where data is actually stored in the drive, and data to be updated. 2. The stacked disk array device according to claim 1, further comprising an address table configured by a logical address of a parity associated with the parity and a parity drive number in which the parity is stored. 6. The stacked disk array device according to claim 1, wherein a set of two or more stacked drives is connected to the RAID controller. 7. A plurality of RAIDs are provided on the RAID controller.
7. The stacked disk array device according to claim 6, wherein a plurality of sets of holes are provided to connect the towers. 8. The stacked disk array according to claim 6, wherein when a set of two or more stacked drives is connected to the RAID controller, a disk array is formed for each set of drives. apparatus. 9. When a set of two or more stacked drives is connected to the RAID controller, a RAID level and the number of drives are changed for each set of drives to form a disk array. The laminated disk array device according to claim 6 or 8. 10. A hole for inserting pins of a plurality of sets of drives is formed on the upper surface of the RAID controller for connecting a set of two or more stacked drives to the RAID controller. The stacked disk array device according to claim 1 or 6. 11. The stacked disk according to claim 6, wherein one RAID tower has two or more sets of buses each including a pin and a hole of a drive in the RAID tower and a hole of a RAID controller. Array device. 12. A RAID tower is provided with two or more sets of buses each including a pin and a hole of a drive in the RAID tower and a hole of a RAID controller, and each drive in the RAID tower is of a usable one. The stacked disk array device according to claim 11, wherein a bus is used.